DATA REDUCTION VERSUS FEATURE SELECTION IN APPLICATION OF DAILY MAXIMUM POWER LOAD FORECASTING

Load forecasting task of small energetic region is a difficult problem due to high variability of power consumption. The accurate forecast of the power in the next hours is very important from the economic point of view. The most important problems in prediction are the choice of predictor and selection of features. Two methods of features selection was presented – simple selection using of genetic algorithm and dimensionality reduction methods for creating new features from many available measured data. As a predictor the Support Vector Machine working in regression mode (SVR) was chosen. The load forecasting results with SVR are presented and discussed.